The AI value gap
Spending on AI in customer service is rising, but few organisations report the return they expected. The gap sits between deploying a model and running a process that delivers a measurable result.
Read the insightDiese Seite ist noch nicht auf Deutsch verfügbar. Es wird die englische Version angezeigt.
Market insight
Budgets are moving towards AI, but the cost of automated resolution is not falling to zero. Buyers increasingly ask what a process achieves, not how many minutes it uses.
AI is taking a growing share of customer-service budgets that are themselves growing slowly.
Published forecasts expect automated resolution to expand, but they also expect the cost per resolution to rise rather than fall to zero.
As pricing moves towards outcomes, minutes and case counts say less about economic value, and sector conditions such as bank credit policy shape the outcomes that can be achieved.
AI spending by customer-service leaders rose by 38%, while overall service and support budgets rose by 2%.
Gartner, press release, 26 August 2026. 199 leaders, survey April–May 2026; not a forecast of any supplier’s revenue or of the voice-AI market.
43% of organisations believe AI could cut contact-centre costs by 30% or more within three years.
Deloitte, The Future of Service: The Age of Intelligent Experience, 16 February 2026. Expectation, not realised savings.
40% of organisations already used agentic AI, and leaders planned to devote 30% of their service budget to AI.
Genesys, 2026 State of Customer Experience. Vendor survey of 5,811 consumers and 1,560 leaders.
Gartner forecasts that agentic AI will autonomously resolve 80% of common customer-service issues by 2029, with a potential 30% reduction in operating costs.
Gartner, press release, 5 March 2025. Forecast; “common issues”, not all contacts.
Gartner forecasts that the generative-AI cost per resolution in customer service will exceed USD 3 by 2030, above the cost of many B2C offshore human agents.
Gartner, press release, 26 January 2026. Forecast; not a forecast of model prices or of any supplier’s costs.
Forrester predicts that rising AI costs will force B2C leaders to scale back about a third of their AI initiatives.
Forrester, Predictions 2027 (B2C marketing, CX and digital), 21 October 2026. Forecast; mainly B2C marketing.
In the third quarter of 2026 banks in Poland significantly tightened their criteria for consumer credit for the first time in ten quarters, despite continued growth in demand for these products.
National Bank of Poland, survey published November 2026, as cited in Vee’s Q3 2026 report. Poland; bank lending survey.
US medical practices named patient no-shows (27%), online scheduling (24%), phone access (22%) and wait times (21%) as patient-access priorities for 2026.
MGMA Stat, Patient access priorities for 2026, 10 December 2025. 236 US practices; priorities, not no-show rates.
45% of practices named insurance and prior-authorisation calls among their most time-intensive phone tasks; 31% named appointment scheduling.
MGMA Stat, Phones are still a bottleneck costing medical practices time, 11 March 2026. 294 responses; sponsored poll; US.
Vee's interpretation. The analysis below is our own reading of the sources above, not a finding of the organisations cited.
When automation is cheap to start but not free to run, the unit that matters is the result: a resolved case, a kept appointment, a payment promise that is honoured. Pricing and reporting follow that unit.
In outcome-linked models, the same number of minutes or cases can correspond to very different economic value, depending on the type of process, the result achieved, attribution rules and the cost of handling exceptions. Shorter conversations that reach the same result can lower minutes without lowering value.
Several billing models usually coexist: fees for activity or time, per-case fees, set-up fees and success-linked components. The choice depends on whether results can be measured reliably and on how much of the outcome the provider can influence.
Gartner’s forecasts point in two directions at once: more issues resolved without a person, and a cost per automated resolution that rises rather than disappears. The economics of a contact process therefore depend on how many contacts reach a result, what each attempt costs, and what exceptions and escalations cost.
Expectations of savings (Deloitte’s 43%) are not the same as realised savings, and the Forrester prediction on AI costs is a reminder that budgets can be withdrawn where returns are unclear.
Consumer finance: where the final outcome of a sales process depends on the lender’s credit decision, tighter credit criteria reduce conversion at later stages regardless of the quality of contact, and change the economics of the process. Receivables monitoring has different operating and economic characteristics from sales.
Healthcare: MGMA’s US polls place no-shows and phone workload among the top access problems for practices, which is where appointment confirmation and rescheduling processes are measured.
Insurance, telecommunications and energy: we hold no sector-specific cost or pricing research for these sectors; the general budget and forecast evidence above applies.
Gartner’s forecasts of 80% autonomous resolution of common issues by 2029 and a generative-AI cost per resolution above USD 3 by 2030. Forecasts.
Forrester’s 2027 prediction that about a third of B2C AI initiatives will be scaled back because of rising costs. Forecast.
Subsequent National Bank of Poland surveys of lending conditions, to see whether the tightening of consumer-credit criteria reported for Q3 2026 continues.
Vee’s new format for presenting operating data (ESPI 11/2026) and the review of accounting policy and presentation of financial information included in the 2027 work programme (Q3 2026 report).
In ESPI 11/2026 (15 September 2026) Vee ended the regular publication of case volumes and minutes of processed conversations in their previous form, explaining that, as the value of settlement increasingly depends on process effectiveness – results, conversion or reach – comparable volumes of cases or minutes may correspond to significantly different economic value. The Q3 2026 report (letter, pp. 10–11) describes a reduction in the scale of some domestic processes related to the sale of credit products amid the tightening of bank credit policy, citing the NBP survey, the parallel development of early receivables-monitoring processes, and billing models that take effectiveness into account alongside traditional fees. The Q2 2026 report (letter, p. 5) stated that the KPI Incentive model would be implemented in Austria from September 2026, with part of the economics of cooperation linked directly to the effectiveness achieved by Vee, and that an analogous solution was being prepared for Germany. The 2025 consolidated annual report describes the move from billing based solely on call time towards models based on the case, quality and business effect.
Market data on this page come from third-party publications and are attributed to their authors. They describe the market, not Vee’s results. Forecasts are the publishers’ own and are not forecasts by Vee S.A. Figures were checked against the publishers’ pages on the date shown.
Spending on AI in customer service is rising, but few organisations report the return they expected. The gap sits between deploying a model and running a process that delivers a measurable result.
Read the insightIn large organisations, AI meets existing telephony, CRM, collection and policy systems. Most barriers reported by enterprises are about connecting to those systems safely.
Read the insightAutomation is accepted when it solves the issue and when a person is reachable if it does not. Enterprises are redesigning human roles rather than removing them.
Read the insightOrganisations record almost every call, yet analyse only a small fraction. Process data is where measurable improvement comes from.
Read the insightIn the European Union, transparency in interactions with AI is now a legal obligation. Regulated sectors expect controls, documentation and clear disclosure that a customer is speaking with an AI system.
Read the insight